DATA VOYAGER ®

Logistic Regression has been used to determine whether a piece of text reflects positive or negative sentiment. Could modification of this technique be used to predict the probability of how well a patient would respond to a drug?

As a thought experiment, could Linear Regression be modified to determine whether a drug regimen would work for a patient (e.g. chemotherapy). With “sentiment analysis”, you match the words in a review with the number of stars in the associated rating (1 or 2 stars is a negative rating; 4 or 5 stars is a positive rating). Next, you convert the review into a matrix (really a sparse matrix) of the counts of the number of times each word appears in each review. Using this data and the associated result (i.e. rating), you can train the model to predict what combination of words reflect the associated sentiment. With large amounts of data, statistical gradient descent works faster and might be applied to such a problem.

Now imagine, if you had for each patient that was given a treatment, the RNA Seq data as to which genes get turned on and in what relative amounts, and what was the outcome of that treatment. Using this data, could you create a logistic regression model to predict whether a treatment could be effective for that patient? The nice thing about Logistic Regression is that not only do you get a prediction of the classification of data, but you also get a probability of that prediction.

The Human Genome Project has made available a massive database of information on the blueprint of human life.  A flood of data in the various fields of biology has led to a systems approach whose goal is to integrate biology, mathematics, and engineering in its efforts to understand and utilize the massive amounts of “information on genes, proteins, cellular dynamics, and organisms’ responses to … the environment” [1]. A deluge of information is also being collected in fields such as physics, meteorology, economics and a host of other disciplines.  The next step is to integrate this data into models and simulations that will allow us to explore, understand, and make predictions at the systems level.  Object Oriented Simulation [OOS] is the natural approach to utilize this data to both model and build simulations at the systems level.  It is an approach that is equally applicable to the diverse fields of biology including neuroscience, epidemiology, and forestry; and also to the field of economics such as the study of market behavior.  But what is OOS?

Read more …
 

October 28,2020

If we look at the Global Financial Crisis (from 10/9/2007 thru 3/5/2009), many would ascribe a major causative factor to the level of mortgage debt in the US. By the first quarter of 2008 (08:Q1), it was 73.66%  of total household debt. Looking at the trend in seriously delinquent debt, that is debt that is 90 days or more delinquent, for the time period of the Global Financial Crisis, the percent of Mortgage debt that was seriously delinquent rose from 4.71% at the start of the period (08:Q1) to a high of 7.39% at the end of the period (09:Q1). It would continue to climb for two more quarters (09:Q3) when it would reach 8.35%, the highest it would be for the entire 17+ years of observation. If, by contrast, we look at what those figures are today (20:Q2) for Mortgage debt, we can see that the trends is much lower and more stable. As a percent of total household debt, Mortgage debt is now 68.53%, and only 1.08% of total Mortgage debt is seriously delinquent.

So where are the future minefields of debt in our economy?

From our observations above, it would seem that trends in percent of total household debt for each loan type (Fig. 1) are somewhat of a leading indicator and trends in seriously delinquent debt for each loan type (Fig. 2) are more of a lagging indicator. Looking at these trends in the percent of total household debt by loan type (Fig. 1), we can see that Student debt and Auto loans appear to be the fastest growing percent of household debt.  Student debt has grown to 11% of total household debt and Auto loans is 9%. After Mortgage debt, they are the largest percent of household debt. And whereas Mortgage debt is now stable, Student debt and Auto loans have been increasing as a percent of Total household debt which is now $14.27 Trillion.

Looking at the trends in seriously delinquent debt for each loan type (Fig. 2), the trend in the percent of Student debt that is seriously delinquent would appear to be the most serious concern. True, in the first and second quarter of 2020, it would appear that this trend in delinquency went down for Student loans. However, this decline in delinquencies was due to the fact that the majority of outstanding federal student loans are covered by CARES Act administrative forbearances. Furthermore, student loans, unlike most other loans (e.g. auto, mortgage, and most credit card debt) cannot be eliminated thru bankruptcy proceedings.

Conclusion

In conclusion, due to the trends in the percent of total household debt that student loans constitute, the percent of student loan debt that is or will be seriously delinquent when the forbearance ends, and total dollar balance of this debt, as well as the nature of this debt, I believe that the next debt minefield be student loans.

Source of Data:

Federal Reserve Bank of New York.  2003 1st Quarter thru 2020 2nd Quarter, 
New York Fed Consumer Credit Panel/Equifax

https://www.newyorkfed.org/microeconomics/hhdc/background.html
https://www.newyorkfed.org/medialibrary/interactives/householdcredit/data/pdf/hhdc_2020q2.pdf

Definitions:

Mortgage                                                                                                                         include all mortgage installment loans, including first mortgages and home equity installment loans (HEL), both of which are closed-end loans.

HE Revolving                                                                                                                    home equity loans with a revolving line of credit where the borrower can choose when and how often to borrow up to an updated credit limit.

Auto Loan                                                                                                                         loans taken out to purchase a car, including leases, provided by automobile dealers and automobile financing companies.

Credit Card                                                                                                                       revolving accounts for banks, bankcard companies, national credit card companies, credit unions and savings & loan associations.

Student Loan                                                                                                                    loans to finance educational expenses provided by banks, credit unions and other financial institutions as well as federal and state governments.

Other                                                                                                                                includes Consumer Finance (sales financing, personal loans) and Retail (clothing, grocery, department stores, home furnishings, gas etc.) loans.

September 21, 2020

In “Why Is the Stock Market So Strong When the Economy Is Weak?” (Knowledge@Wharton, Aug 31, 2020), Wharton finance professor Itay Goldstein identifies three (3) factors for the disconnect between Wall Street and Main Street:

1. “The stock market is meant to be forward-looking”
2. “that the Fed started injecting all this money into the market pushed prices up”
3. “the stocks that are doing well – Google, Facebook, Amazon, Microsoft, Netflix – haven’t been hurt that much by the current economic conditions”

In this blog, I would like to focus on the third factor.

If we focus on the S&P 500, we can see just how much influence the largest stocks (defined by market capitalization) has on this index.

The S&P 500 is a good proxy for the stock market because it represents a broad swath of the stock market. The ETF of the S&P 500 that is often used is the SPDR® S&P 500 ETF Trust (symbol: SPY). It is a market capitalization weighted ETF. And by that I mean that the fraction of the index for a company in the index is equal to the market cap of that individual company divided by the sum of the market caps for all the companies in the index. Market Cap is defined as the total number of outstanding shares a company has multiplied by the current price of a single share.

If we look at the top 5 companies (by market capitalization) that make up the S&P 500 index, we can see that they make up over 22% of the market valuation of the index and have an average YTD % return that is nearly 3 times that of the index as a whole.

	% of the index	YTD % return
Apple Inc	6.51%	46%
Microsoft Corp	5.54%	27%
Amazon.com Inc	4.59%	60%
Facebook Inc	2.26%	23%
Alphabet Inc (class A)	1.62%	8%
Alphabet Inc (class B)	1.58%	9%
Total % of the Index:	22.10%
Mean YTD % return		29%
SPY Market Price YTD % return		10%

 

Another way of seeing how much of an oversized influence those large market capitalization stocks have on the index, is to compare an S&P 500 market capitalization weighted index to its equal-weighted counterpart. An equal-weighted index, as its name implies, weights each stock equally in the index. If you had 500 stocks and the total value of the index is $1,000, then you would invest $2 into each stock. An ETF that is the equal-weighted version of the S&P 500 is the Invesco S&P 500® Equal Weight ETF (symbol: RSP).

	Annualized Returns (%)
	1 Year	5 Years	10 Years
RSP (equal weighted) Market Price	-3.40%	6.86%	12.25%
SPY (capitalization weighted) Market Price	7.37%	10.61%	13.84%

 

Finally, if we graph the 12 month returns for the equal-weighted vs the capitalization weighted ETFs for the S&P 500 for 5 years, we can see that the capitalization weighted ETF (SPY) outperformed the equal-weighted ETF (RSP), especially for the last 12 months (ending on 9/9/2020).

In conclusion, the returns that you are seeing are predominantly from the largest capitalization Technology stocks and not necessarily from the stock market as a whole. And while these stocks have propelled the S&P 500 index up in the last 12 months, they could also drag the market down in the next 3 months.

February 7, 2017

In the January 23, 2017 issue of Barron’s, in the 2017 Roundtable, Part 2, Jeffrey Gundlach made the comment, “One of the best indicators of the direction of bond yields is the ratio of copper prices to gold prices.” He went on to elaborate, “Copper is an industrial metal. A higher ratio suggest more manufacturing activity, and that implies an uptick in inflation and yields.”

I thought that it would be interesting to see whether there was a correlation between this ratio and CPI, and between this ratio and 10 year US Treasury rates.

The two metrics which Mr. Greenblatt uses is “Return on Capital” and “Earnings Yield”. In the appendix of “The Little Book That Beats The Market”, he defines these terms as follows:

Return on Capital = EBIT/(Net Working Capital + Net Fixed Assets)

EBIT = earnings before interest and taxes

Earnings Yield = EBIT/Enterprise Value

For more details on Mr. Greenblatt’s methodology, consult his book or his website: www.magicformulainvesting.com

He ranks a list of stocks (e.g. top 1,000 US non-financial, non utility stocks with over $1 billion revenue) by “Return on Capital” and “Earnings Yield” (the largest Return on Capital would be ranked 1, and the largest Earnings Yield would also be ranked 1). And then he adds the two ranks to get a combined rank. Then, for example, the top 30 ranked would be purchased and held for a year.

Comparing the returns for this portfolio (1988 thru 2009) to that of the S&P 500, the portfolio beat the S&P 500 in every year except the following: 1989,1990,1993,1999,2002,2008.

The Periods of Recession in the US during this time period (1988 thru 2009) were:

• July 1990 – March 1991
• March 2001 – November 2001
• December 2007 – June 2009

As we can see, this strategy did better than the market in the 2001 recession, but worse in the 1990 and 2008 recessions.

 

 

January 10, 2017

In Joel Greenblatt’s book, “The Little Book That Beats The Market”, he writes that one could get better than market returns by “buying good companies (ones that have a high return on capital) and ” … “only at bargain prices (at prices that give you a high earnings yield)”. I would like to explore that concept for different time periods (e.g. years, months, etc), and different markets (e.g. expansions, recessions).